Urban ecophysiology: beyond costs, stress and biomarkers

As urban environments expand and rapidly encroach on natural habitats, it is increasingly important to understand the impact of urbanisation on organisms. Urban habitats are characterised by impervious surfaces and buildings, higher levels of chemicals, noise, light at night, increased ambient temperatures and availability of anthropogenic food sources that are often of poor nutritional quality (Isaksson, 2018). Although recent studies have shown that suburban habitats can host a high biodiversity (Batáry et al., 2017), highly urbanised areas show an overall decline in biodiversity (e.g. McKinney and Lockwood, 1999; Shochat et al., 2010), which is projected to decline further as urban habitats are growing (McDonald et al., 2018, 2020). Less obvious is the impact that the urban habitat has on city dwellers (e.g. pigeons, crows, rats and racoons). Thus, a key question in urban ecology is whether current urban-dwelling species will be able to cope with the expanding urban sprawl and its associated environmental changes. How challenging is the urban environment for these species?

Physiology plays a pivotal role in allowing individuals to cope and persist in the urban environment. Urbanisation and its associated anthropogenic factors are predicted to affect physiological systems such as the redox system (leading to oxidative stress; see Glossary; Box 1), inflammation, metabolism and endocrinology (Navara and Nelson, 2007; Isaksson, 2015; Isaksson and Bonier, 2020). The measurable components of these physiological systems include different anti-oxidant enzymes, organic co-factors (such as vitamins), metabolites (e.g. amino acids), proteins (such as cytokines and haptoglobin) and hormones (e.g. corticosterone). These components are regularly used as (bio)markers of effect and/or exposure (Boxes 1 and 2).

By measuring physiological markers, we can gain insight into individual- and species-level sensitivity and resilience to urbanisation, but can also begin to understand future threats of expanding urban sprawl, highlighting potential areas of concern. Conversely, we can gain insight into the potential positive outcomes of conservation efforts. Although studies of physiological markers in an urban context have the potential to be highly valuable, their interpretation is not always straightforward. Take, for example, the two most commonly investigated aspects of physiology in an urban context – oxidative stress and the hypothalamic–pituitary–adrenal (HPA) axis (Box 1) – which show mixed responses to urban life (e.g. Bonier, 2012; Isaksson and Bonier, 2020) (Table 1).

This Commentary aims to go beyond the simplistic heuristic biomarker approach (Box 2), and explore when and why results are not always as predicted. What causes the frequently observed variation in physiology across (and even within) urban environments? Does our current biomarker framework limit future advances in ecophysiology and its importance in evaluating the impact of urbanisation? I argue that this is indeed the case. I do not aim to review all challenges, responses or taxonomic groups, but instead focus on the systems with which I am most familiar (i.e. oxidative stress and vertebrates), in order to highlight general concerns that can be extrapolated to other systems and responses.

The field of urban ecophysiology is relatively young, going back to the early 2000s (e.g. Partecke et al., 2004; Bonier et al., 2007), and the field of urban evolutionary physiology is even younger (reviewed in Isaksson and Bonier, 2020). Most research has been on birds, with a focus on a few European and North American species, such as great tits (Parus major; e.g. Isaksson et al., 2009), house sparrows (Passer domesticus; e.g. Meillère et al., 2015), blackbirds (Turdus merula; e.g. Partecke et al., 2006; Dominoni et al., 2013), house finches (Carpodacus mexicanus; e.g. Giraudeau et al., 2015) and song sparrows (Melospiza melodia; Grunst et al., 2014). However, the field is rapidly growing, and knowledge about other species, taxonomic groups and their physiological responses in different geographical and climatic regions is increasing (e.g. French et al., 2008; Falfushinska, et al., 2008a, 2018b; Radwan et al., 2010; Payne et al., 2012).

Although it has become clear that some aspects of the urban environment are attractive to some organisms (e.g. high abundance of food and nesting sites), the general conception is that, overall, the urban environment is challenging for its inhabitants (e.g. Austin, 1998; Samet et al., 2004; Hartz et al., 2008; Acevedo-Whitehouse and Duffus, 2009; Isaksson, 2010; Koivula and Eeva, 2010). The attractive features of the city can cause some species to preferentially settle in an urban instead of a non-urban habitat, with detrimental effects on fitness, a so-called ‘ecological trap’ (e.g. Demeyrier et al., 2016; Vlaschenko et al., 2019). However, the nature and magnitude of the urban challenge can vary widely, and this variation is not only a result of city size, but also of other abiotic and biotic factors that may vary within and between cities; these are discussed in more detail below.

At the macro-scale, cities are rather homogeneous. Macro-level ecology is clearly valuable for understanding general features of city dwellers in comparison with those from rural areas (e.g. Møller, 2009; Møller et al., 2010). Here, however, I want to highlight the importance of the microscale for understanding the physiological variation across the urban landscape. In fact, when looking at specific urban factors, both within and across cities, cities can be highly environmentally heterogeneous (both spatially and temporally). Take, for example, air pollution. In proximity to highways and roads, air pollution can reach dangerously high levels. Even small cities (∼300,000 inhabitants) can occasionally exceed the recommended threshold for acceptable levels of certain air pollutants (European Environment Agency, 2018; Salmón et al., 2018a). However, cities can improve urban air quality, by allowing more green areas and trees to grow (Pugh et al., 2012). Similar spatial and temporal variation can be found for other urban factors, such as noise and night light pollution (e.g. Sprau et al., 2017).

The effects of temporal and spatial variation in exposure level within a city has been shown for humans, with increased incidence of air pollution-related diseases on peak days and when living on exposed streets (e.g. Oudin et al., 2010). Recently, ecologists have started to either experimentally manipulate or quantify variation in single or multiple urban factors and relate this to physiological markers (e.g. Raap et al., 2016; Casasole et al., 2017; McLay et al., 2017; North et al., 2017; Salmón et al., 2018a). Among these studies, only McLay et al. (2017) link the physiological effects caused by artificial light to measures of fitness. By doing so, the physiological marker can more reliably be used as a biomarker of effect. Also, as physiology can respond more rapidly and directly than life-history traits to short-term environmental perturbations, the spatial and temporal variation in urban factors can be used to disentangle the effects of urban stressors on physiological responses and thereby their potential long-term effects on the population/species.

For wildlife, a ‘challenging’ environment is defined as one that causes physiological responses or outcomes to deviate from the ‘normal’, with ‘normal’ in the context of urbanisation often being nearby population(s) from a non-urban habitat. As discussed above, urbanisation presents a broad range of challenges, thereby also eliciting a broad range of physiological responses. When the challenge is extreme, such as very high pollution levels or lack of an essential micronutrient such as thiamine or calcium, there is no doubt that urbanisation negatively affects wildlife. However, in most cases, the challenges are not acutely toxic, which allows physiological responses to compensate, leading to acclimation (see Glossary) to prevailing conditions (e.g. Kight and Swaddle, 2011; Isaksson, 2015; Isaksson and Bonier, 2020). This complicates the answer to the question ‘when and why is an urban environment challenging?’.

In many cases, the interpretation of whether an urban environment is challenging, based on measured physiological responses, has been simplistic and deterministic, with the physiological responses being interpreted as either ‘good’ or ‘bad’ for the urban dwellers. This can lead to false conclusions about the impact of urbanisation. Take, for example, corticosterone, which, until recently, has been used as a measure of stress (MacDougall-Shackleton et al., 2019). Thus, when studied in the urban context, the general hypothesis has been that corticosterone levels should be higher in city dwellers compared with those of their non-urban conspecifics. According to this hypothesis, a higher corticosterone level would then trigger an appropriate behavioural response to the urban challenge (Bonier, 2012). However, the results have been highly mixed, and no general association between urbanisation and individual ‘stress’ has been established (Bonier, 2012). As discussed by MacDougall-Shackleton et al. (2019), these apparently differing responses relate to the simplistic and deterministic view of this group of hormones. Glucocorticoids are, in fact, multi-faceted – they are involved in many physiological functions (see also Jimeno et al., 2018). By no longer assuming a direct association of these hormones with ‘stress’, and realising that responses can acclimate or evolve in response to urbanisation, we will achieve a better understanding of their role in urban-dwelling organisms. This also applies to a wider range of physiological responses, such as oxidative stress and metabolism.

Apart from the physiological complexity, there is also variation in degree of urbanisation across studies. Thus, a lack of an effect in one study may simply be a result of lower degree of urbanisation compared with another study with stronger physiological effects (for oxidative stress, see Table 1). Although it is becoming more common to include standardised urbanisation indexes in the description of study sites (Seress et al., 2014), this should be common practice so that the physiological responses can be compared across studies and evaluated based on degree of urbanisation.

The biomarker approach (Box 2) in ecophysiology originates from the fields of medicine and ecotoxicology. This approach has, to some extent, also been valuable in urbanisation research, allowing researchers to identify specific challenges, and determine the physiological cost and health status of the population (Isaksson, 2010; Murray et al., 2019). However, as discussed above, the physiological responses in the wild show high variability (e.g. Falfushinska et al., 2008a, 2008b; Koivula and Eeva, 2010; Bonier, 2012; Abdel-Halim et al., 2013; Gillis et al., 2014; Herrera-Dueñas et al., 2014; Salmón et al., 2018a, 2018b; see also Table 1), which is why we need to broaden our view and to explore the underlying causes for this variability.

The redox system and its response to oxidative stress (Box 1) is the primary aspect of physiology that is used as a source of biomarkers for environmental stress (see Table 1). In the urban environment, oxidative stress can change in response to, for example, increased pollution and ambient temperatures, and poorer food quality. The intriguing aspect of oxidative stress is its association with cellular ageing and dysfunction; it thus provides a link to a physiological cost, which can be manifested as reduced fitness (e.g. Metcalf and Alonso-Alvarez, 2010; Isaksson et al., 2011). Yet the data that support these links in an urban context are still scarce (but see Lucas and French, 2012).

In recent years, it has been emphasised that one must measure many components (biomarkers) of oxidative stress in order to fully understand the oxidative stress status of an individual (e.g. Beaulieu and Costantini, 2014; Speakman et al., 2015). However, even when doing so, the results can be challenging to interpret (e.g. if individual markers show opposite responses; Table 1). Hence, the use of the strict biomarker definition when using components of the redox system to evaluate whether an urban environment is challenging may only be reliable and valuable when the urban challenge is very high and toxic (Box 1). This may also apply to other physiological responses. Moving beyond physiological determinism and instead focusing on context-dependent constraints in physiology and the processes leading to a specific response will be much more rewarding. I will expand on this below.

The way in which species, populations and individuals physiologically respond to urbanisation can be influenced by, for example, phylogenetic constraints, time since colonisation, life history, life stage, and current and previous diseases, along with nutrition and other environmental factors. Thus, extrapolations about the significance of a certain marker from one context to another can be misleading. To answer the question of who finds the urban environment challenging, at the physiological level, will rely on detailed knowledge about the species/population and the urban environment, but also on the capacity for physiological plasticity and robustness (see Glossary).

Within an individual, physiology is highly plastic and responsive to environmental perturbations. Some physiological responses occur immediately (within seconds to hours), whereas others require days to weeks to adjust. Likewise, some responses rapidly subside, whereas others are maintained for a longer time. Maintaining a strong response over an extended period is often associated with a cost, such as chronic inflammation and/or alternative resource allocation (i.e. resources are withdrawn from other functions).

Usually, urban factors are not acutely detrimental (but see Calderón-Garcidueñas et al., 2014, 2015), and thus are manageable in the short term (Isaksson and Bonier, 2020). On longer time scales, they can, however, be detrimental. Acclimation and other kinds of physiological plasticity are likely to play an important role in reducing the short- and long-term effects, thereby increasing species resilience towards urbanisation. Despite the high interest in plasticity, studies of urban physiology (like most urban ecology studies) mainly use effect size analyses and summary statistics between urban and non-urban sites, which obscure the potential to explore the underlying variance across different contexts (see Terblanche and Hoffmann, 2020). Cross-fostering, common garden and translocation studies would also be useful for understanding both the capacities for plasticity across naïve and pre-exposed individuals and populations, and the underlying mechanism(s) behind the variation (i.e. genetic or ‘non-genetic’) (Costantini et al., 2014; Romero-Haro and Alonso-Alvarez, 2015; Isaksson and Bonier, 2020). Indeed, a direct plastic response of the anti-oxidant enzyme superoxide dismutase was detected using a cross-fostering design, where urban-raised chicks had an increased activity of the enzyme regardless of nest or habitat of origin (Salmón et al., 2018b). This plastic response was, however, not sufficient to prevent generation of lipid peroxides in late-hatched urban broods.

Furthermore, the distinction between baseline and acute level responses is common practice in some ecophysiological fields, such as endocrine ecology and immuno-ecology (Quaye, 2008; Bonier et al., 2009; Matson et al., 2012; Hegemann et al., 2013; Vermeulen et al., 2016; Ouyang et al., 2017). This distinction is unfortunately rarely seen in fields such as oxidative stress (but see Costantini et al., 2014; Raap et al., 2016). Instead, the levels of both anti-oxidants and oxidative damage are usually measured at a single time point and without an experimental challenge to elicit a plastic response. Within-individual changes can reveal important information about the range and capacity of responses across time and contexts. Although oxidative damage is not directly a physiological response, but an outcome, it was recently shown that even oxidative damage to lipids can vary rapidly within an individual (Eikenaar et al., 2020). Thus, it would be highly valuable to also view oxidative damage as a variable trait, and to explore patterns of clearance and regeneration of oxidative damage over time to establish when it actually becomes a threat to fitness.

In addition to displaying physiological plasticity, physiological markers and systems are also highly integrated. For example, hormones can affect metabolism, which in turn affects the production of reactive oxygen species (ROS), thereby affecting components in the anti-oxidant system. This integration is a crucial part of physiological function and robustness (see also Cohen et al., 2012; Martin and Cohen, 2015). However, ecophysiological studies – and not just in urban contexts – rarely consider this.

Physiological integration (and thereby robustness) might explain the lack of a linear relationship between a specific marker and environmental stress. It has been suggested that this might account for difficulties in revealing physiological trade-offs in the wild (Metcalf and Alonso-Alvarez, 2010; Williams, 2018). Physiological integration means that the ‘value’ of a physiological marker is not necessarily constant across contexts and species. To exemplify this, I will use a bank analogy used by Cohen et al. (2017) in a theoretical paper about ‘currencies’ in life-history trade-offs, but will expand it to the present context. Before presenting the analogy, I will summarise three main points about physiological trade-offs (i.e. ‘currencies’) in an ecological context. First, we know that there are multiple possible ‘currencies’, such as dietary anti-oxidants and calories, that can be allocated to different functions. Second, the value of these currencies is context dependent and non-linear (i.e. they display diminishing marginal returns and are sometimes non-monotonic: it is possible to have too much; e.g. Isaksson and Bonier, 2020). This means that the ‘values’ of a marker vary across species, environments, populations and territories, for example. Third, the value of the currencies is not necessarily equal for different functions, such as reproduction versus somatic maintenance, and some level of each currency is necessary (as can be seen from nutrient shortages causing health problems even when there are sufficient calories).

Now to the analogy (see Fig. 1). Imagine that a bank offers reward points (RPs) as one form of currency. In this context the RPs can be carotenoids (AOX1) – a group of dietary substances that has multiple biological functions (Britton, 1995; Svensson and Wong, 2011). AOX1 can be invested in either ‘savings’ (i.e. stored in tissues for later use) or in ‘spending’ (i.e. directly used). In addition to RPs, there is also a second currency: ‘money’ (i.e. resources such as calories). If the value of the money and RPs are equal (or at least similar) to different functions such as survival and reproduction, there are almost an unlimited number of ways that you could allocate the RPs and money to still arrive at your optimum allocation. However, if the money has different values in savings versus spending, the presence of two currencies will allow more optimised solutions. Now, if the relative values in savings and spending are different in different populations and the different currencies do not have equal market values (i.e. there are different resource availabilities, which can also vary over time), then your allocation strategy will become more important for the final outcome. Additionally, different banks (territories) provide different RPs (AOX1 is instead AOX2, -3, -4 or a combination of two or more). In other words, the dietary availability of certain nutrients (RPs) differs in different territories (banks), which means that the optimum strategy (phenotype) is reached in different ways in different territories either within or across urban environments (Fig. 1). This analogy is in line with current evolutionary theories that promote the importance of the ecological context for understanding life-history strategies (e.g. Williams, 2018).

Physiological compartmentalisation is also part of physiological robustness. Different tissues are differentially affected by urban stressors. For example, inhaled air pollutants most strongly affect the lungs and respiratory tract, whereas ingested toxins may most affect the liver. This will influence the level of a physiological response in certain tissues, but also means that different stressors will affect different markers. Not surprisingly, blood is the most commonly used tissue in ecophysiology, as its sampling is non-lethal. However, when assessing markers of oxidative stress and inflammation, blood is probably the tissue that is most integrated and potentially the most variable in its response (Speakman et al., 2015; Isaksson, 2015; Ouyang et al., 2018; Costantini, 2019; Isaksson and Bonier, 2020). For example, different cytokines have different half-lives, which means that any point measurement indicates both current and previous responses, which could either obscure, under- or over-estimate an association with fitness. Hence, when using blood compared with other tissues, it is even more important to understand and investigate the causes for individual and population level variation, to be able to assess its suitability as a biomarker (Box 2).

Not only do environmental conditions trigger physiological responses: they can also be part of the response. For example, many different nutrients such as trace metals, essential fatty acids and amino acids are required for multiple cellular functions and structures, and their environmental availability can influence an individual's response to urban factors, leading to city-specific responses.

The high availability of anthropogenic food is a key factor allowing city fauna to thrive (e.g. Murray et al., 2016; Kumar et al., 2018). Luring wild animals to feed on these anthropogenic sources can, however, lead to a less diverse and nutritious diet (Birnie-Gauvin et al., 2017; Shulte-Hostedde et al., 2018), affecting their physiology. For example, city dwellers tend to have increased cholesterol, decreased dietary anti-oxidants and altered fatty acid composition compared with rural individuals (Andersson et al., 2015; Isaksson et al., 2017; Townsend et al., 2019). These are conditions that, in humans, are associated with increased risk of developing cancer, metabolic and cardiovascular diseases, and hence reduced lifespan. Although cancer has recently been highlighted as a concern also for wild animals (Madsen et al., 2017), urban great tits show no difference in the expression of anti-cancer genes compared with their rural conspecifics, which the authors argue could represent a lag in the evolutionary response to a highly oncogenic environment (Giraudeau et al., 2020). However, it could also be that anti-cancer genes are expressed in other tissues than blood. Moreover, it is not always the absolute concentration of a certain macro- or micronutrient that is important, but the balance between them (Simpson and Raubenheimer, 2014). For example, in humans, the modern Western diet increases the relative intake of omega-6 to omega-3 polyunsaturated fatty acids (PUFAs), which triggers a downstream pro-inflammatory response, leading to a state of chronic inflammation (Calder, 2011). Similar dietary changes have been shown for urban birds (Isaksson et al., 2017); this could further enhance inflammatory responses in urban wildlife. However, it could also stimulate evolutionary changes that reduce the physiological value of nutrients such as omega-3 PUFAs in urban habitats over time, such that these become less essential in urban-dwelling individuals. This is not unrealistic; for example, hummingbirds have evolved to survive on an extreme sugar diet, with very few other macro- or micronutrients.

Ecological studies rarely detect or screen for diseases linked to malnutrition. Instead, several nutritional markers such as carotenoids, cholesterol, Mead acid and omega-6/omega-3 ratio have been associated with ‘body condition’ (Andersson et al., 2015; Townsend et al., 2019), a state used to infer survival prospects and ‘health’ in a broad sense, which – under natural conditions – is highly reliable (Peig and Green, 2009). The high availability of anthropogenic food can reduce the immediate risk of starvation by increasing the body reserves of urban-dwelling animals; thus, a positive association between anthropogenic food resources and body condition could be predicted (reviewed in Jones and Reynolds, 2008). However, in the future, we also need to consider whether high intake of anthropogenic food, and thereby high body condition, might make wildlife more susceptible to inflammatory diseases and environmental stress. Indeed, in urban black sparrowhawks (Accipiter melanoleucus) relatively low body condition is associated with increased survival (Sumasgutner et al., 2016; Tate and Amar, 2017).

Furthermore, other micronutrients such as minerals do not show a consistently lower or higher availability when comparing urban versus rural habitats, but instead show a high overall variation in environmental availability. Thus, independently of urbanisation, environmental differences in nutrient availability can contribute to inconsistencies in physiological responses in urban habitats. For example, the availability of calcium is known to vary across environments, and it limits uptake and accumulation of heavy metals such as arsenic, lead and cadmium (Dauwe et al., 2006; Scheuhammer, 1996). Thus, a deficiency in calcium can have toxic effects, whereas calcium supplementation can mitigate the potential negative effects of living in an otherwise highly polluted environment. In addition, calcium intake seems to mitigate the inhibitory effect that heavy metals have on the anti-oxidant system; thus, variation in calcium intake across urban sites and across urbanisation gradients (see Glossary) can influence the anti-oxidant response (Sánchez-Virosta et al., 2019).

Thus, macro- and micronutrients have potentially constraining, mitigating and enhancing effects on physiological responses, thereby influencing resilience/sensitivity towards urbanisation. These effects have been overlooked in ecophysiology, and require more attention.

The first and most urgent future direction in urban ecophysiology is to link physiological components to fitness under different urban conditions (e.g. temporal or spatial differences, nutrient availability or pollution) using experiments or long-term data sets. This will allow a better understanding of the underlying causes for physiological variability or consistency under different conditions, and thereby its potential use as a ‘biomarker of effect’. Well-designed experiments will also be able to disentangle the underlying mechanisms of acclimation (or other types of plasticity), heritability or genetic adaptation (see Glossary), or will determine whether the responses are coping strategies or even maladaptive, allowing us to understand the impact of urbanisation. These kinds of experimental studies are still rare (but see Lucas and French, 2012; McLay et al., 2017; Salmón et al., 2018b; Tuzun and Stoks, 2020).

Another future direction to explore physiological integration and robustness can be to use ‘omics’ (i.e. proteomics, metabolomics, transcriptomics; see Box 3) approaches, allowing over-represented pathways, functions and metabolites to be identified (e.g. Suarez and Moyes, 2012; Watson et al., 2017; Gouveia et al., 2019). This knowledge can then allow us to target key physiological markers or systems that are affected by urbanisation, which can potentially lead to faster progress relative to measuring single physiological markers without prior knowledge of which physiological components are affected by urbanisation.

Furthermore, in order to increase our understanding of urban ecophysiology at multiple levels and scales (from species-level to cellular responses, from macro- to micro-ecological processes, and from current to future responses) the field would benefit greatly from multi-disciplinary projects and open access data-sharing. This would allow us to disentangle the effects of different pollutants (either alone or in combination), abiotic factors (e.g. ambient temperature, soil acidity and water stress) and biotic factors (e.g. food availability and pathogen transmission), while maintaining the ecological relevance.

Urban environments are home to a range of different species. Although some of these species thrive, they face a number of physiological challenges, such as increased pollution and ambient temperatures. The traditional biomarker approach to detect physiological costs of urban life has gained mixed results. In this Commentary, I argue that unless the urban challenge is severe, the biomarker approach could lead to inaccurate conclusions about the physiological responses to urban challenges due to heterogeneity in the abiotic and biotic factors that are integrated into the physiological response. Furthermore, the fundamental properties of a biomarker – namely that their levels should be consistent for a certain state, as well as across time and contexts, e.g. season and life-stage – are often violated or ignored, resulting in the misuse and misinterpretation of physiological markers in response to urbanisation.

Physiological complexity and variability may seem daunting; however, rather than dismissing physiology as an ecological tool to estimate the impact of a given stressor, a conceptual paradigm shift is needed, where physiological, ecological and evolutionary processes are integrated in much greater detail. For this to happen, we need to move beyond the biomarker concept and physiological determinism, and instead focus on context-specific causes for variation, physiological plasticity and robustness and evolutionary processes within, as well as across, urban environments. Only then can ecophysiology claim to increase our understanding of the underlying causes of environmental stress, and accurately estimate the resilience of species, populations and individuals to future anthropogenic developments. Finally, I wish to finish this Commentary by rephrasing Dobzhansky's old saying from 1973: Nothing in urban ecophysiology makes sense except in its ecological and evolutionary context (Dobzhansky, 1973). I hope that this Commentary will inspire (especially young) researchers to take a more holistic and integrated view of physiology and provide a framework for the design of future studies to gain a better understanding of the physiological responses and thereby the impact of urbanisation. Overall, I think that urban ecophysiology has an important role to play for understanding species resilience to urbanisation and that it has an exciting future ahead.

I thank Alan Cohen and three anonymous reviewers for highly valuable and insightful feedback that greatly improved this paper.